For people who grow, transfer, measure and build: crystal growers, CVD and MBE practitioners, spectroscopists, microscopists and device fabricators. It gives concrete conditions, the signals that diagnose what you made, and the ways experiments go wrong without anyone noticing.
This page is written for specialists. For a plain-language introduction, start with the basics.
You are reading as a theoretician. This is the experiment track; the theory track is the one written for you.
You are reading as an engineer. This is the experiment track; the engineering track is the one written for you.
Mechanical exfoliation in four steps. Tape lifts thin stacks off a layered crystal, repeated peeling thins them, and pressing the tape onto an oxidised silicon wafer leaves flakes behind. Thin-film interference in the oxide makes flakes of different thickness look different, so even a single layer can be found by eye before being confirmed spectroscopically. Exfoliation in the glossary →
Growth methods
11 ways to make the material, with the conditions that matter, how far each one scales, and the ways each one fails.
Flakes on a chip
the highest-quality material, for physics and prototype devices
A batch process giving centimetre-sized boules in days. Industry grows bulk Bi2Te3 thermoelectric ingots this way, but for 2D work it supplies crystals to exfoliate, not wafers.
Batches of tens of grams demonstrated in the laboratory; fluoride handling is the main constraint.
What each method gives you, from its own entry below: the form the material comes in, not a ranking. Select a method for how it works and where it fails.
Mechanical exfoliation (including metal-assisted)
How it works
Adhesive tape or a stamp overcomes the weak between layers of a bulk crystal; are pressed onto a and located optically.
Read moreShow less
Typical conditions
Room temperature; low-residue tape; SiO2/Si with 90 nm or 285 nm oxide. Oxygen-plasma cleaning of the substrate and peeling from a warm substrate (~100 °C) increase flake size and yield. Gold-assisted variants use a freshly deposited Au layer that bonds strongly to atoms.
Good for
The highest-quality material for physics and prototype devices, from any layered crystal that can be grown in bulk.
How far it scales
None in the industrial sense. Metal-assisted reaches millimetre-to-centimetre but adds a metal-removal step.
Where it goes wrong
Low, random yield of monolayers
Tape and on flakes
Flake quality capped by the source crystal – bulk defects come along
Folds, wrinkles and cracks from aggressive peeling
Practical tips
Clean substrates by O2 plasma and exfoliate onto warm substrates to improve adhesion
Peel slowly: a few deliberate peels give larger flakes than many rapid ones
For crystals (black phosphorus, CrI3, NbSe2, WTe2), exfoliate and encapsulate entirely inside a glovebox
Liquid-phase and electrochemical exfoliation
How it works
Sonication or high shear in a solvent whose surface energy matches the crystal separates the layers.
Read moreShow less
Alternatively, ions are electrochemically to push layers apart before gentle agitation.
Typical conditions
NMP, IPA/water or aqueous surfactant (e.g. sodium cholate); bath or tip sonication for hours, or shear mixing; centrifugation cascades to select flake size and thickness. Electrochemical: cathodic intercalation of tetraalkylammonium cations for , anodic sulfate intercalation for graphite.
Good for
, , coatings, printed electronics, and and materials.
How far it scales
Excellent – kilograms to tonnes for graphene-related materials. Quality control, not volume, is the challenge.
Where it goes wrong
Broad distributions of flake size and thickness
Solvent and surfactant residues – NMP, boiling at 202 °C, is hard to remove
Sonication-induced defects and fragmentation
Oxidation of reactive materials during processing
Practical tips
Use liquid cascade centrifugation to split one dispersion into size-selected fractions
Estimate flake size and thickness from UV–vis extinction spectra with published metrics, but calibrate against statistics
Electrochemical intercalation with bulky organic cations gives larger, more uniform TMDC monolayers than sonication
Chemical vapour transport (CVT)
How it works
A sealed, evacuated quartz ampoule holds the elements or a pre-reacted charge plus a , in a temperature gradient.
Read moreShow less
The agent forms volatile species with the solid at the source, which diffuse along the ampoule and decompose at the sink, depositing crystals. The direction of transport follows the sign of the transport reaction’s enthalpy: carry material from hot to cold, exothermic reactions from cold to hot.
Typical conditions
Quartz ampoules evacuated to ~10−3–10−5 mbar and flame-sealed. Iodine is the most common agent for TMDCs, typically a few mg per cm3 of ampoule volume; Br2, TeCl4, SeCl4 or NH4Cl are alternatives. Source and sink temperatures are usually in the ~750–1,050 °C range with a gradient of ~50–100 K, over one to several weeks.
Good for
High-quality bulk for exfoliation: TMDCs, magnets (CrI3, CrSBr), MPS3 and many other layered compounds. Also the fastest way to explore new compositions, , dopants and .
How far it scales
A batch process giving millimetre-to-centimetre crystals over weeks – ideal for research supply, not for manufacturing.
Where it goes wrong
Transport agent incorporated into the crystal – iodine or chlorine as dopant or intercalant, changing
Many small crystals instead of a few large ones when nucleation at the sink is uncontrolled
Wrong or mixed polytypes (2H/3R MoS2, 1T/2H TaS2, 2H/1T′ MoTe2) depending on temperature and cooling rate
Non- and from imbalanced loading or chalcogen loss
Ampoule failure from excessive chalcogen or halogen pressure at high temperature
Oxides or oxyhalides from moisture or oxygen in poorly dried precursors or quartz
Practical tips
Acid-clean and bake out quartz; dry precursors; when evacuating, keep the transport agent cold so it does not sublime into the pump
Pre-react the elements into a charge before transport to avoid runaway chalcogen pressure
Clean the sink by briefly reversing the gradient before growth, so crystals nucleate on a bare wall
Keep the gradient small for fewer, larger crystals; a large gradient speeds transport but multiplies nucleation
Estimate chalcogen and halogen partial pressures for your ampoule volume before sealing – and use thicker-walled quartz if in doubt
Choose the cooling protocol deliberately: quenching versus slow cooling selects metastable versus stable polytypes
Report agent, loading in mg/cm3, both temperatures, duration, ampoule dimensions and cooling – together they set crystal quality and residual doping
Flux and self-flux growth
How it works
Crystals grow slowly from a molten solvent during controlled cooling – excess chalcogen () or a low-melting metal or salt – without any transport agent.
Read moreShow less
Typical conditions
Sealed ampoules with excess Se or Te for selenides and tellurides; slow cooling at a few kelvin per hour or slower; flux removed by hot centrifugation or sublimation. The hBN in most high-quality devices is grown at high pressure and temperature, about 4.5 GPa and 1,500 °C, from a barium boron nitride solvent (Watanabe and Taniguchi, NIMS); at atmospheric pressure, Ni–Cr or Fe–Cr fluxes also give large, high-purity crystals.
Good for
Crystals with markedly lower densities than typical vapour-transport crystals – important for optics and high- devices.
How far it scales
Batch and slow, but yields some of the cleanest research crystals available.
Where it goes wrong
Flux inclusions trapped inside crystals
Very long growth times, weeks to months, for large crystals
Self-flux is hazardous for sulfides because of sulfur’s vapour pressure
Practical tips
Decant the flux at temperature through a quartz-wool filter in an inverted ampoule
Anneal crystals under chalcogen vapour to heal vacancies
Benchmark each batch with a defect-sensitive measurement such as low-temperature linewidth or defect counts
Bridgman and melt growth
How it works
The melt is sealed in an ampoule with a pointed tip and lowered slowly out of the hot zone of a furnace through a temperature gradient.
Read moreShow less
The tip freezes first, and the grain that nucleates there – ideally only one – grows into a single crystal as the solidification front climbs through the melt. A stoichiometric melt works only for compounds that melt congruently; others, such as InSe, are grown from a melt of adjusted composition.
Typical conditions
Elements or a pre-reacted charge, sealed under vacuum in a quartz ampoule with a conical or capillary tip, often lined with pyrolytic carbon. The hot zone sits a few tens of kelvin above the melting point – 585 °C for Bi2Te3, 706 °C for Bi2Se3, 960 °C for GaSe – and the ampoule is lowered through a gradient of roughly 10–30 K per cm at 0.5–2 mm per hour, then cooled slowly over a day or more.
Good for
Centimetre-sized single crystals of layered compounds that melt, such as InSe, GaSe, Bi2Se3, Bi2Te3, SnSe and PbI2; a common source of exfoliated InSe and of and crystals.
How far it scales
A batch process giving centimetre-sized boules in days. Industry grows bulk Bi2Te3 thermoelectric ingots this way, but for 2D work it supplies crystals to exfoliate, not wafers.
Where it goes wrong
Several grains instead of one when more than one nucleus survives at the tip, often with twins or low-angle boundaries
Composition drifting along the boule, because the growing crystal rejects the excess of one element into the remaining melt
Volatile Se, Te or I lost to the free volume, leaving vacancies and antisite defects that dope the crystal – Bi2Se3 comes out n-type this way
Cracks from thermal stress or from the crystal sticking to the quartz
and mixed polytypes in compounds with several stackings, such as GaSe and InSe
Practical tips
Coat the inside of the ampoule with pyrolytic carbon so the crystal does not stick to the quartz and crack on cooling
Use a sharp or capillary tip so that only one grain survives to seed the boule
Lower slowly – about 1 mm per hour – and keep the solid–liquid interface flat or slightly convex towards the melt
Grow InSe from a slightly indium-rich melt, since it does not melt congruently
Cleave pieces from the start, middle and end of the boule and check orientation, grain count and composition before exfoliating
Powder-source CVD of TMDC monolayers
How it works
Metal oxide (MoO3, WO3) and chalcogen powders are vaporised in a tube furnace; metal suboxides react with chalcogen vapour on the substrate to nucleate and grow monolayer domains.
Read moreShow less
Typical conditions
Substrate at ~650–850 °C; sulfur heated separately to ~150–250 °C; Ar or Ar/H2 carrier (H2 is needed for selenides); SiO2/Si, sapphire or mica substrates; NaCl or KI promoters lower the growth temperature and enlarge domains; growth for ~10–30 minutes.
Good for
Fast access to monolayer MoS2, WS2, MoSe2, WSe2 and their alloys; lateral and vertical by sequential growth; growth-mechanism studies.
How far it scales
Centimetre-scale in tube furnaces; not uniform enough for wafer production.
Where it goes wrong
Poor run-to-run reproducibility from precursor depletion and position-dependent vapour concentration
Alkali contamination (Na, K) from salt promoters that changes electronic properties
Oxygen incorporation and chalcogen vacancies
Multilayer nucleation at domain centres and random domain orientation on amorphous SiO2
Practical tips
Fix precursor mass, boat positions and the onset of sulfur heating precisely – when chalcogen arrives relative to oxide evaporation controls nucleation
Use c-plane sapphire for aligned domains; controlled miscut can make them unidirectional
Map PL and across whole domains – bright edges or dark centres reveal defects and gradients
CVD of graphene and hBN on metal foils
How it works
Carbon or boron–nitrogen precursors decompose on a catalytic metal surface; the low solubility of carbon in copper makes graphene growth largely self-limiting at one layer.
Read moreShow less
Typical conditions
Graphene: copper foil annealed in H2 near 1,000–1,070 °C, then CH4/H2 at low or atmospheric pressure. hBN: ammonia borane or borazine on Cu, Pt or Fe–Ni at ~1,000 °C.
Good for
Large-area monolayer graphene and hBN, including single-crystal films on Cu(111) or by suppressed nucleation.
How far it scales
Roll-to-roll and wafer-scale growth demonstrated.
Where it goes wrong
Grain boundaries and multilayer islands
Copper roughness and impurities nucleating defects
Wrinkles from thermal-expansion mismatch on cooling
Transfer damage that dominates the final film quality
Practical tips
Electropolish and anneal copper to reduce nucleation density
Characterise on the growth foil before transfer – briefly oxidising the copper in air reveals graphene domains – so you know which defects come from growth
MOCVD of wafer-scale TMDCs
How it works
Volatile metal–organic precursors (e.g. Mo(CO)6, W(CO)6) and chalcogen sources (H2S, diethyl sulfide, dimethyl selenide) are supplied at low, controlled partial pressures, giving uniform layer-by-layer growth over whole wafers.
Read moreShow less
Typical conditions
Roughly 500–950 °C depending on precursor and target quality; very low precursor partial pressures and growth times of hours for monolayer control; sapphire or SiO2/Si wafers.
Good for
Uniform wafer-scale monolayers for device integration; industrial research on 200–300 mm platforms.
How far it scales
The most industrially relevant route; equipment makers offer reactors specifically for .
Where it goes wrong
Carbon contamination from metal–organic ligands
Small grains at low growth temperatures
or highly regulated precursors (H2S, H2Se)
Best-quality growth temperatures exceed
Practical tips
Control background water and oxygen tightly – they change nucleation density and doping
Add H2 to suppress carbon incorporation, balanced against its of the film
Map thickness and optical uniformity across the whole wafer, not at a single point
MBE and van der Waals epitaxy
How it works
Elemental beams in ultra-high vacuum condense on a heated substrate.
Read moreShow less
For layered materials the weak substrate interaction relaxes the need for lattice matching – .
Typical conditions
Base pressure around 10−10 mbar; chalcogen-rich flux ratios; substrates such as graphene on SiC, graphite, mica, sapphire or Au(111); growth rates of order a monolayer per hour; in-situ , often coupled to STM or .
Good for
Clean samples for surface spectroscopy, that exist only on substrates, magnetic and thin films, and heterostructures with atomically sharp interfaces.
How far it scales
Wafer-scale coverage is possible, but TMDC crystallinity generally trails the best CVD and MOCVD films.
Where it goes wrong
Small grains and rotational domains on weakly interacting substrates
Chalcogen deficiency and in TMDCs
A narrow window between adatom desorption and three-dimensional islanding
Practical tips
Calibrate flux ratios and substrate temperature against RHEED streak sharpness
Cap air-sensitive films in situ with Se, Te or AlOx before breaking vacuum
Try two-step growth: low-temperature nucleation followed by annealing under chalcogen flux
Etching and delamination of MXenes
How it works
The A-element layers of a are removed selectively in fluoride-containing acid; the resulting multilayer carbide is intercalated and delaminated into single flakes.
Read moreShow less
Typical conditions
A widely used recipe: LiF dissolved in 9 M HCl, Ti3AlC2 added slowly, stirred at ~35 °C for ~24 h; repeated washing by centrifugation until the pH approaches neutral; delamination by hand shaking or brief sonication in deoxygenated water.
Good for
, water-dispersible flakes for films, inks and electrodes.
How far it scales
Batches of tens of grams demonstrated in the laboratory; fluoride handling is the main constraint.
Where it goes wrong
Incomplete etching (residual MAX phase) or over-etching (TiO2 formation)
Low delamination yield from insufficient washing
Oxidation of dispersions during storage
Hydrofluoric acid exposure
Practical tips
Start from well-crystallised MAX powder – MAX quality carries through to flake stability
A dark, almost black colloidal supernatant after washing signals successful delamination
Never work with HF or LiF/HCl without HF-specific training and calcium gluconate gel at hand
Post-growth conversion and Janus synthesis
How it works
Convert an existing film or monolayer chemically: sulfurise or selenise deposited metal or metal-oxide films, or replace one chalcogen face of a finished monolayer to create a layer.
Read moreShow less
Typical conditions
Metal-film chalcogenisation: typically ~600–1,000 °C in sulfur vapour or H2S; PtSe2 by selenisation of Pt at ~400 °C. Janus MoSSe: hydrogen-plasma stripping of the top sulfur layer followed by selenisation at a few hundred °C, controlled sulfurisation of MoSe2 near ~800 °C, or room-temperature plasma-assisted replacement.
Good for
Fab-compatible routes that start from deposited metal films, and the only practical way to make Janus MXY monolayers.
How far it scales
Metal-film conversion scales to wafers; Janus conversion is still at flake or centimetre scale.
Where it goes wrong
Polycrystalline films with small grains
Incomplete or non-uniform conversion – alloyed rather than truly Janus layers
Vacancies and etching from plasma damage
Loss of monolayer integrity at high temperature
Practical tips
Calibrate plasma power and exposure time on sacrificial samples, checking Raman after each step to catch the point where the top layer is stripped
Grow intentionally alloyed MoS2xSe2(1−x) reference samples and compare their PL and Raman with your Janus candidates
Map conversion across the whole flake – edges and grain boundaries convert differently from interiors
Transfer and stacking
7 techniques for moving a layer off what it grew on and assembling it into a stack. Most of what limits a finished heterostructure happens here.
Dry pick-up, the usual way to build an encapsulated stack from exfoliated flakes. A polymer stamp picks up the top hBN, then uses it to lift the graphene, so the graphene only ever touches hBN. The stack is laid down on the bottom hBN slowly on a heated stage, so the contact front sweeps trapped contamination out of the interface, and the PC film is dissolved. The failure modes are under “Dry pick-up with polymer stamps” below.
Wet polymer (PMMA) transfer
Used forMoving CVD graphene or TMDC films from copper, sapphire or SiO2 onto a target substrate
What can go wrong
PMMA residue that dopes and scatters carriers
Cracks and tears during etching or scooping
Etchant contamination (Fe from FeCl3; Na or K from KOH)
Water trapped under the film
What to do
Etch copper with ammonium persulfate rather than FeCl3 to avoid iron residues
Rinse through several deionised-water baths, with a dilute HCl step to remove metal ions
Reduce residue by annealing in Ar/H2 or by using sacrificial interlayers
Dry and bake slowly so trapped water can escape
Electrochemical bubbling transfer
Used forReusable growth foils (Pt, Cu) and transfers that must avoid etchant contamination
What can go wrong
Mechanical damage from hydrogen bubbles
Incomplete delamination along grain boundaries
What to do
Use low current density and dilute NaOH electrolyte
Support the film with a stiff polymer frame during delamination
Dry pick-up with polymer stamps (PC, PPC on PDMS)
Used forAssembling van der Waals heterostructures from exfoliated flakes;
What can go wrong
Bubbles of trapped hydrocarbons and water
Polymer residue if the stamp melts onto the stack
Misalignment or tearing of flakes
What to do
Pick up hBN first so that graphene or TMDCs only ever touch hBN
Laminate slowly at elevated temperature so the contact front sweeps contamination out of the interface (the approach)
Dissolve PC in chloroform, then rinse in acetone and isopropanol
Tear-and-stack for twisted bilayers
Used forTwisted homobilayers with a chosen angle
What can go wrong
relaxation during heating
Local angle inhomogeneity and heterostrain
What to do
Keep stacking temperatures low and avoid post-annealing
Use a precise rotation stage and verify the resulting angle, for example by or lattice-resolved imaging
Deterministic transfer with viscoelastic PDMS stamps
Used forPlacing individual flakes onto pre-patterned electrodes, cavities, or strain platforms
What can go wrong
Oligomer residue from PDMS
Strain or cracking during stamp release
What to do
Release slowly while gently heating the target substrate
Pre-clean PDMS, or add a thin PPC film to it to reduce residues
Interface cleaning after assembly
Used forBefore device fabrication or spectroscopy of stacks
What can go wrong
Annealing can rotate twisted stacks or degrade air-sensitive layers
Contact-mode AFM sweeping can tear flakes
What to do
Sweep contamination out of active regions with a contact-mode AFM tip at low force
Ar/H2 annealing coalesces bubbles – but not for twisted or air-sensitive stacks
Confirm cleanliness with AFM topography and dark-field optical imaging
Wafer-scale transfer and bonding
Used forMoving MOCVD films onto -compatible target wafers
What can go wrong
Cracks, wrinkles and incomplete coverage over 100–300 mm
Polymer or adhesive residues not tolerated in fabs
Metal contamination
What to do
Use -grade adhesives and wafer bonders rather than hand transfer
Quantify coverage and defect density by automated optical inspection
What each measurement tells you
16 techniques: the question each one answers, the signal to look for, where it stops being reliable, and the mistake it most often invites.
Which technique answers which question, as read from each one’s “Tells you” line below. Where several answer the same question, let them check each other: a layer count read from optical contrast, for one, needs calibrating against AFM or Raman on the same substrate. Select a technique for what to look for and where it stops being reliable.
Optical contrast microscopy
Tells youRapid location of thin flakes and an estimate of layer number
Read moreShow less
What to look forContrast depends on oxide thickness and wavelength; monolayer graphene on ~300 nm SiO2 shows roughly 5–10% contrast in green light
Where it stopsCalibration-dependent; weak for hBN and other materials; blind to stacking and defects
Common mistakeAssigning layer number from colour without calibrating against AFM or Raman on the same substrate and illumination
Raman spectroscopy
Tells youLayer number, strain, doping, defects, stacking order and crystal phase
Read moreShow less
What to look forGraphene: G ~1,580 cm−1, 2D ~2,680 cm−1 and defect-activated D ~1,350 cm−1 (532 nm). MoS2: –A′1 separation ~18–20 cm−1 for a monolayer, ~25 cm−1 in bulk. Low-frequency shear and below ~50 cm−1 count layers and reveal stacking. hBN: E2g ~1,366 cm−1.
Where it stopsStrain and doping shift the same peaks; laser heating shifts peaks and damages samples; resonance changes intensities with excitation wavelength
Common mistakeUsing too much laser power on monolayers, and reading peak shifts as strain or doping without separating the two
Photoluminescence (PL)
Tells youDirect-gap monolayers, and energies, defect-bound excitons, strain, doping and optical quality
Read moreShow less
What to look forMonolayer A excitons near 1.85–1.9 eV (MoS2), ~2.0 eV (WS2), ~1.57 eV (MoSe2) and ~1.65 eV (WSe2); linewidths of a few meV at cryogenic temperatures in hBN-encapsulated samples; low-energy defect bands at low temperature
Where it stopsIntensity alone is a poor quality metric, because substrate interference and doping change it; ensemble spectra hide spatial variation
Common mistakeComparing PL intensities between samples on different substrates or oxide thicknesses
Second-harmonic generation (SHG)
Tells youCrystal orientation, , layer parity, twist angle, grain boundaries and polar order
Read moreShow less
What to look forStrong SHG from odd-layer 2H TMDCs and none from even layers; sixfold polarisation patterns reveal the armchair axes; interference between stacked layers yields the twist angle
Where it stopsRequires pulsed lasers; resonance enhancement complicates quantitative comparisons
Common mistakeTreating SHG intensity as a thickness measure without accounting for resonance and substrate effects
Atomic force microscopy (AFM)
Tells youTopography, thickness, bubbles, wrinkles and surface contamination
Read moreShow less
What to look forIntrinsic layer thickness ~0.34 nm for graphene and ~0.65 nm for MoS2, but flake-to-substrate steps often read 0.6–1 nm or more
Where it stopsAdsorbed water, residues and tip–sample interactions distort absolute heights; tapping-mode heights can depend on imaging parameters
Common mistakeCounting layers from one flake-to-substrate step instead of from steps between layers within the same flake
Kelvin probe and conductive AFM
Tells youSurface potential, , doping variations and local conductivity
Read moreShow less
What to look forContrast between layer numbers, grain boundaries and doped regions; work functions when the tip is calibrated against a reference such as graphite
Where it stopsSurface adsorbates dominate in air, and the tip work function drifts
Common mistakeReporting absolute work functions measured in ambient air without tip calibration
Scanning tunnelling microscopy and spectroscopy (STM/STS)
Tells youAtomic lattice, point defects, , local and gaps
Read moreShow less
What to look forDefect types and charge states; ; peaks and correlated gaps in moiré systems
Where it stopsNeeds conductive substrates or gated devices; tip-induced in semiconductors
Common mistakeAssigning defect identity from topography alone – vacancies and substitutional atoms can look alike
ARPES and micro/nano-ARPES
Tells you, , band alignment and doping
Read moreShow less
What to look forWhether the valence-band maximum sits at K or Γ distinguishes monolayer from bilayer TMDCs; flat bands in moiré stacks; -resolved ARPES resolves
Where it stopsProbes occupied states only; exfoliated flakes need micrometre-scale beams; samples must be clean and electrically grounded
Common mistakeIgnoring substrate photoemission and sample charging, or final-state effects in few-layer samples
X-ray photoelectron spectroscopy (XPS)
Tells youComposition, oxidation states, stoichiometry and contamination
Read moreShow less
What to look forMo 3d5/2 near 229–229.5 eV for MoS2 versus ~232.5–233 eV for MoO3; S 2p3/2 near 162 eV
Where it stopsAverages over tens to hundreds of micrometres; charge referencing to adventitious carbon can mislead
Common mistakeFitting peaks without constraining spin–orbit doublet splittings and area ratios
TEM, STEM-HAADF and 4D-STEM
Tells youAtomic structure, defects, grain boundaries, stacking, strain maps, and Janus face assignment in cross-section
Read moreShow less
What to look forZ-contrast distinguishes S from Se and Mo from W; 4D- maps strain and local twist angle
Where it stops creates the very defects being counted; fields of view are small
Common mistakeImaging beam-sensitive TMDCs at 200–300 kV instead of 60–80 kV with controlled dose
EELS and EDS in the electron microscope
Tells youElemental maps, bonding, and – with monochromated EELS – excitons and vibrations
Read moreShow less
What to look forSingle-atom dopant detection; excitonic and features in monochromated spectra
Where it stopsDose, signal-to-noise and quantification errors; EDS of monolayers is dominated by the support
Common mistakeQuantifying monolayer composition by EDS on SiN or carbon supports without careful background correction
XRD and grazing-incidence scattering (GIWAXS)
Tells youPhase, polytype, , crystallinity of bulk crystals and in-plane orientation of films
Read moreShow less
What to look forBasal (00l) reflections give the interlayer spacing; rocking curves assess mosaicity; reveals film texture
Where it stopsWeak signals from monolayers; bulk-averaged
Common mistakeConcluding phase purity from strong (00l) reflections alone – different polytypes share the same basal spacing
Low-temperature magnetotransport
Tells youCarrier density, mobility, , and states, and magnetism
Read moreShow less
What to look forHall slope for density; Shubnikov–de Haas oscillations for effective mass and quantum mobility; Landau fans on several contact pairs for homogeneity
Where it stopsContacts dominate two-terminal measurements; inhomogeneity mixes regions
Common mistakeReporting two-terminal field-effect mobility that includes , without saying which mobility it is
Magneto-optics (MOKE, RMCD)
Tells you and in micrometre-sized flakes
Read moreShow less
What to look for or reflective magnetic loops versus thickness – for example layered antiferromagnetic steps in few-layer CrI3
Where it stopsNeeds a cryostat with optical access; weak or absent signals for antiferromagnets
Common mistakeNeglecting Faraday rotation in the objective and cryostat windows under field – always measure a non-magnetic reference
Ultrafast and terahertz spectroscopy
Tells youCarrier and exciton dynamics, and contact-free conductivity
Read moreShow less
What to look forInterlayer charge transfer on timescales in heterobilayers; terahertz photoconductivity of large-area films
Where it stopsInterpretation depends on models; high fluences introduce nonlinear effects
Common mistakeUsing pump fluences that drive exciton–exciton annihilation and then attributing the decay to intrinsic processes
Near-field optics (s-SNOM) and cathodoluminescence
Tells youNanoscale response and emission: polaritons, stacking domains and emitter locations
Read moreShow less
What to look forPhonon polaritons in hBN and α-MoO3; contrast between stacking domains in twisted and rhombohedral graphene
Where it stopsSpecialised instruments; electron beams can charge or damage samples
Common mistakeInterpreting near-field amplitude without the phase or a model of tip–sample coupling
Making devices
8 steps between a good flake and a measurement you can trust, and what each step can hide.
Where each part of making a device acts, numbered as the entries below: an hBN-encapsulated channel with a local graphite gate over the silicon back gate, contacts on its ends, and a resist residue sealed in under the top layer.
Evaporated Ti/Au, Cr/Au or Ni contacts are standard but damage the lattice and pin the . Semimetal contacts (Bi, Sb), gentle low-energy evaporation of In or Au, transferred prefabricated electrodes, and graphene contacts reduce damage and Schottky barriers.
Watch outDamage and interface reactions depend on deposition rate, chamber pressure and substrate heating – contact resistance can differ by orders of magnitude between nominally identical recipes.
Lithography and resist residues
PMMA electron-beam lithography is standard; resist left on channels degrades mobility and dopes the material.
Watch outResidues after development and lift-off are invisible in an optical microscope – only AFM before and after shows them.
Gate dielectrics on inert surfaces
nucleates poorly on -free basal planes, leaving pinholes and islands. Seed layers (thin oxidised metal films or molecular seeds), mild surface pre-treatments, or hBN as the dielectric are the usual solutions.
Watch outAggressive plasma or ozone pre-treatments create defects, and seed layers shift and .
Gating schemes
Global silicon are simple but weak; local graphite or metal gates with hBN dielectrics give clean dual-gated devices; ionic-liquid and ion-gel gates reach carrier densities above 1014 cm−2, but the gate voltage can only be changed while the electrolyte is liquid, above about 200 K, and is then frozen in on cooling.
Watch outIonic gating can intercalate into or react with the sample – check reversibility to separate electrostatic from electrochemical doping.
Encapsulation
hBN encapsulation for the highest quality; ALD oxides or polymer for larger areas.
Watch outEncapsulation also seals in whatever was already on the surface, so clean before you encapsulate.
Thermal budget
Post-fabrication anneals in forming gas or vacuum lower contact resistance and remove residues, but also change doping, create vacancies and can rotate twisted stacks.
Watch outRecord every thermal step, including resist bakes – for air-sensitive or metastable samples they all count.
Hall bars and four-probe geometries
Etch Hall bars (fluorine-based plasmas for hBN stacks, O2 plasma for graphene) to extract intrinsic mobility and carrier density.
Watch outEtched TMDC edges are defect-rich and can conduct in parallel; account for edge contributions.
Statistics and yield
Measure many devices: distributions of mobility, threshold voltage, hysteresis and contact resistance say more than a champion device.
Watch outHysteresis measured at different sweep rates, ranges or atmospheres is not comparable – state all three.
Why results do not reproduce
10 reasons the same experiment gives two answers – in two labs, or in the same lab two months apart.
Use them as a checklist for the result in front of you: tick each one you have dealt with. The ticks stay in this browser only.
The crystal source varies
How it bites youThe same compound from two suppliers or two growth runs can differ in point-defect density by orders of magnitude.
What to doReport supplier and batch or full growth parameters, and benchmark each batch with a defect-sensitive measurement.
Samples age in air
How it bites youMonolayers oxidise and collect hydrocarbons within hours to days, so properties drift between the first and last measurement.
What to doLog time in air between fabrication and measurement, store samples in vacuum or inert gas, and remeasure a control sample.
The laser changes the sample
How it bites youRaman and PL lasers heat, photo-oxidise and dope samples during mapping.
What to doRun a power series on a sacrificial spot and report laser power, spot size and objective.
Substrate interference
How it bites youOxide thickness modulates Raman, PL and intensities through thin-film interference.
What to doState substrate and oxide thickness, and compare intensities only on identical substrates.
Contacts masquerade as material properties
How it bites youTwo-terminal mobilities and record on-currents can be mostly contact effects.
What to doExtract contact resistance with or measurements and state which mobility is reported.
Hysteresis from adsorbates and traps
How it bites youThreshold voltages shift with sweep direction, rate and humidity.
What to doMeasure in vacuum after pumping, use pulsed measurements, and report sweep conditions.
Hidden dopants
How it bites youSodium or potassium from , iodine from vapour transport and iron from etchants all dope samples.
What to doCheck with , ToF-SIMS or elemental analysis, and always report promoters and transport agents.
Champion-device reporting
How it bites youBest-device numbers without distributions make results impossible to compare or reproduce.
What to doReport device counts and full distributions, and share raw data.
Look-alike structures
How it bites youRandom alloys can mimic Janus layers, and phase mixtures can mimic pure polytypes, in single optical measurements.
What to doCombine several structural probes – Raman, SHG, STEM, XPS – before assigning a structure.
Characterisation damage
How it bites youElectron and ion beams create the defects being quantified.
What to doUse low accelerating voltages, dose series and fresh areas, and report dose.
Facilities you can apply to
8 shared instruments beyond what a single group owns. All of them take external proposals unless their note says otherwise; read each centre’s own call for deadlines and eligibility.
German national centre for advanced electron microscopy
User programme
Searches to follow
15 standing arXiv searches for this track. They are part of what fills this site’s news feed, and each link opens the live feed from the arXiv API for a feed reader to subscribe to; the query itself is written out so you can change it.
Janus monolayers
Synthesis and characterisation of Janus TMDCs
abs:"Janus" AND (abs:"monolayer" OR abs:"MoSSe" OR abs:"WSSe" OR abs:"transition metal dichalcogenide")
cat:cond-mat.mtrl-sci AND (abs:"chemical vapor deposition" OR abs:"chemical vapour deposition" OR abs:MOCVD) AND (abs:monolayer OR abs:"two-dimensional")
Bulk crystal growth by vapour transport, flux and Bridgman methods for layered and 2D materials – restricted to condensed-matter categories because growth-method terms alone also match astrophysics and fluid dynamics
(cat:cond-mat.mtrl-sci OR cat:cond-mat.str-el OR cat:cond-mat.mes-hall) AND (abs:"chemical vapor transport" OR abs:"chemical vapour transport" OR abs:"flux growth" OR abs:"self-flux" OR abs:"flux method" OR abs:Bridgman OR abs:"single crystals were grown") AND (abs:"van der Waals" OR abs:dichalcogenide OR abs:"two-dimensional" OR abs:monolayer OR abs:exfoliated OR abs:"few-layer")
Air-sensitive 2D materials present fundamental challenges for device integration. Encapsulation is required to preserve intrinsic properties, yet conventional strategies require complicated fabrication workflows and fail with thicker flakes. We demonstrate that electron-beam (e-beam) evaporated aluminum oxide (AlOx)…
Spectrally isolated, electrically driven emission from localized states in two-dimensional semiconductors remains challenging in scalable planar devices. Here, we report narrow-line-width alternating-current electroluminescence (ACEL) from localized states in monolayer WSe2 using a lithographically defined sub-5 nm…
Transition metal ditellurides exhibit structural and electronic properties distinct from their selenide and sulfide counterparts, including the stabilization of multiple (CDW) phases. Within the niobium compound family, notable differences in structure are also observed: bulk NbTe2 crystallizes in…
We theoretically investigate the optical response of the WSe2 monolayer vertically stacked on twisted bilayer graphene (tBG) under electrostatic doping. In this heterostructure, the doped moiré superlattice of tBG generates a spatially modulated electrostatic potential that couples to the constituents…
We numerically study exciton- coupling in a hybrid structure composed of a period-doubled Si3N4 photonic crystal slab and an hBN-encapsulated MoSe2 monolayer. Period doubling folds quasi-guided modes into the light cone and produces spectrally separated photonic branches whose radiative character is controlled by…
13 papers worth reading in full, and why each one is here.
Making graphene visibleBlake et al. · Applied Physics Letters 91, 063124 (2007)cited by 1,799The interference physics behind optical flake identification – still the basis of every flake search.
Chemical Vapor Transport ReactionsBinnewies, Glaum, Schmidt & Schmidt · De Gruyter (book) (2012)cited by 159The standard text on transport thermodynamics, agents and practical ampoule work.